I know there are 25,478,852 tools that say they will build everything for you.

I’ve tried several tools, and none of them have what I want, so I decided to build it myself because I’m trying something a bit different.

My goal is to keep the developer in control of all decisions. I want to bring a “team” to help with game creation, not create everything for you.

Most importantly, I want to build something where a new Hermes, OpenClaw, or any new model would only be additive, not competing and forcing you to choose one.

I’m not trying to build a “one-shot” creation tool. My approach is that you work your normal job and discuss things with the agents as if they were colleagues.

At the end of the day, you can playtest, prepare a plan for the next day, or leave it running overnight to review and polish.

I don’t know if this is the correct way or not, but I think I’m onto something.

I would love to know your opinion about it.

This week, the game itself moved forward in a much better direction.

One of the main improvements was creating different bullet types based on elements. Instead of having one generic projectile, bullets can now behave differently depending on their element type. That opened the door for more interesting combat and better enemy variety.

The project is now much more data-driven, which made it easier to create different types of enemies with different behaviors without rebuilding everything from scratch. This was one of the biggest wins of the week.

We also added some basic player stats, like health and stamina, along with a double jump. These are small systems, but they already make the prototype feel more like an actual game and less like a simple test scene.

We also created the foundation for data-driven boss mechanics. The idea is to have bosses share the same basic structure, but still allow each one to have different movement patterns, attack sets, and combat behaviors. It is still early, but the direction feels right.

The hardest part for me right now is introducing proper assets. I don’t think the framework is strong enough there yet. But for prototyping, testing mechanics, building abilities, shaping the game loop, and giving the developer a solid playground to work from, it is working really well.

This will probably be the last incremental update for the FPS prototype for now. Next week, I want to take what I learned over the last three weeks and try to build a top-down shooter with two abilities.

That should be a good test to see how much of the framework can transfer to a different type of game.

https://arthur0n.itch.io/xf3d-fps-demo-3